A real-time cooperative driving framework for un-signalized intersection

Connected Autonomous Vehicles (CAVs) have gradually gained a larger market share in recent years. But at present, there are still significant limitations, especially the inadequacy of decision-making abilities. Leveraging the cooperative driving capabilities of CAVs holds promise to solve this problem. However, as the number of vehicles increases, the computation efficiency often becomes another challenge. Therefore, the potential game and conflict search tree are combined in this paper to establish a real-time cooperative driving framework for CAVs. Instead of assuming players’ decision-making follow given patterns, the potential game implicitly generates such behaviors by optimizing the potential function. Additionally, the cooperation problem is decomposed into several sub-problems by conflict search tree and solved in a distributed manner. Results show that the proposed framework exhibits good computation efficiency in multiple experiments and is capable of ensuring system safety.